[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-tool-pairs-reasoning-with-scans-to-find-tricky-lesions":10,"sections":40},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},11352,"ai-tool-pairs-reasoning-with-scans-to-find-tricky-lesions","AI Tool Pairs Reasoning With Scans to Find Tricky Lesions","A new framework teaches AI models to explain their reasoning before segmenting complex medical lesions, outperforming prior methods on a new benchmark.","Researchers built an AI system that explains its reasoning before it draws the boundaries of a lesion in a medical scan.\n\nThe team introduced two things: ComLesion-14K, a benchmark of complex lesion cases paired with step-by-step reasoning chains, and CORE-Seg, a framework that links a language-reasoning model to a pixel-level segmentation tool through a module they call a Semantic-Guided Prompt Adapter. Training runs in two stages, supervised fine-tuning followed by a reinforcement-learning stage called GRPO, with a reward system designed to keep useful feedback flowing even on hard cases. On the new benchmark, CORE-Seg posted a mean Dice score (a standard measure of overlap between predicted and actual lesion boundaries) of 37.06%, which the researchers say is 14.89% higher than the second-best baseline. It also cut its failure rate to 18.42%.\n\nSegmentation tools are good at drawing boundaries but bad at explaining why. General-purpose multimodal AI models are the opposite: broad common sense, little specialized visual judgment. Combining both in one pipeline, rather than bolting a chatbot onto a segmentation model, is the more interesting bet here, and the new benchmark may end up mattering as much as the model, since reasoning-labeled lesion data has been scarce.\n\nA 14.89% edge on a brand-new benchmark that the same team built is worth noting, not popping champagne over; the real test is whether this reasoning holds up on lesions nobody wrote a training chain for.","[\"ai\",\"medical-imaging\",\"reinforcement-learning\",\"healthcare\"]","2026-10-12T04:00:00.000Z","2026-10-12T09:50:23.931Z","2026-10-12T09:50:28.731Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"The source states the Dice score is '14.89% higher than the second-best baseline' without specifying whether that's percentage points or a relative percentage increase, but the draft confidently converts it to 'about 15 percentage points ahead' — rewrite using the source's own phrasing (or clarify units) instead of assuming an interpretation the source doesn't support.","resolved","ai",[30,32,33,34],"medical-imaging","reinforcement-learning","healthcare",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.05911",0,{"sections":41},[42,46,51,56,61,66,71,76,81,86,91,96,101,106],{"name":43,"slug":30,"count":44,"latest_published_at":45},"AI",6967,"2026-10-12T15:20:00.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Security","security",954,"2026-10-12T14:20:37.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Policy","policy",496,"2026-10-11T20:06:43.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Deals","deals",492,"2026-10-12T14:55:53.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Hardware","hardware",237,"2026-10-12T13:24:45.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Science","science",196,"2026-10-11T15:34:40.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Consumer Tech","consumer-tech",192,"2026-10-12T14:07:13.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Startups","startups",121,"2026-10-12T13:22:34.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Software","software",115,"2026-10-11T07:34:45.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Dev Tools","dev-tools",111,"2026-10-12T11:22:24.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"General","general",71,"2026-10-12T16:10:40.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"Gaming","gaming",59,"2026-10-09T11:43:43.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"Reviews","reviews",35,"2026-10-12T14:00:00.000Z",{"name":107,"slug":108,"count":109,"latest_published_at":110},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]